|Tukanoan languages in northwest Amazonia. h/t Wiki.|
|A map of the proposed Tupi-Carib-Gê family, which helpfully shows the approximate distributions of all three families. These maps, as Michael Heckenberger (among others) has remarked, show solid colours where the local distribution is actually much more mixed and difficult to interpret. h/t Wiki.|
These proposed macro languages families typically involve established families that are found in geographic proximity to one another, which means that any claim of genetic relatedness has to overcome the possibility that their similarities are the result of language contact over a sustained period. For example:
- Panoan and Tacanan languages are spoken within a relatively small area of the Upper Amazon on the edges of the Andes and must have had significant impacts on one another as they developed over time, so unravelling the reality of the connections between them is tricky (there is an Arawak wedge between some of the Pano-Tacanan languages, but this seems to be a relatively recent, albeit pre-Columbian, occurrence).
- The languages included in the Macro-Gê proposal happen to be right inside or adjacent to Gê-speaking territory, meaning that while they may indeed be related, it is hard to ascertain whether the relationships are genetic or areal.
- Quechua and Aymara have been spoken next to one another in a region inter-connected by trade, migration, and shared religious belief for thousands of years, and almost the entirety of Quechua- and Aymara-speaking land was included in the Inka empire in pre-Columbian times.
- Chibchan and the languages proposed as sub-branches of Macro-Chibchan are spoken right next to one another in northern Colombia through to Costa Rica.
- There is evidence of borrowing between Tupi-Guarani (a Tupian sub-family) and Carib, which has implications for understanding the origins of both language families. Assuming this is correct, it undermines the Gê-Tupi-Carib proposal slightly, as Carib is clearly not a sub-family of Tupian and the similarities between them might be due to this contact. But technically it doesn't negate the proposal in its entirety, especially as the purported relationship is claimed on the basis of grammatical rather than lexical similarities.
- Arawak languages have been in contact with several language families, including Carib, Tukanoan (spoken in northwest Amazonia), Panoan, Tupian, Gê, and even Quechuan, but since they are so widely spread, distributed themselves relatively recently (perhaps with the last 2-3000 years), and are so closely associated with certain socio-cultural features, determining their relationships isn't so complex.
It doesn't help, either, that many of these languages are dying out, that speakers are adopting and being affected by the Indo-European languages of Spanish and Portuguese, and that insufficient research was done in the past. The impact of the post-Columbian world was downright apocalyptic in some (not all, but certainly some) areas, with disease and deliberate genocide killing enormous numbers of native people across the continent - meaning, among other horrific things, that lots of native languages were lost or had their speakers merge with other communities.
All of this must have obscured, and continues to obscure, our understanding of the relationships between South American languages. Besides the fragmented geography, there is no particularly reason why South American languages should be so difficult to make sense of. We've got to assume that a large part of the problem consists of sampling problems and language extinction.
There are also plenty of isolates (languages with no or very few relatives) and unclassified languages. As one of the main purposes of historical linguistics is the ability to reconstruct past population movements and attempt to make sense of archaeological and ethnographic data, these isolates are in some respects less interesting to prehistorians. But their presence is suggestive of other features of South American prehistory, including the tendency towards preserving languages that would otherwise have been eliminated.